We propose to develop X-ray area detectors for macromolecular crystallography, based on photoelectric absorption in crystalline silicon wafers. In such a sensor, diffracted X rays will be captured directly in thick, high-resistivity, silicon, micromachined by inductively-coupled plasma etching. The etching process we will use cuts clean, sharply-defined features in silicon without creating defects in neighboring material. A 2D raster of etched holes will be filled with conducting polycrystalline silicon ("poly")-half of which will be n type and the other half p type. These poly columns will serve as electrodes of a reverse-biased diode array for generating drift fields and for electron capture. Each square 150 mum pixel will have one n- and one p-type electrode, and will be bonded with a conducting bump to a corresponding input in a CMOS readout chip, located behind the sensor. Plasma etching will also be used to cut out each silicon sensor. The smooth, defect-free edges thus produced will be lined with a conducting poly electrode, making the sensor wafer active to the very edge. Overlapping arrays of these sensors will therefore have no dead gaps. The CMOS electronics will be fabricated by standard (0.25 mum technology) photolithographic processing but will be designed to have a large safety factor against radiation damage. We will design, fabricate, debug, test, and characterize several modules during the first stage of the proposed project, along with the off-chip electronics, physical packaging, and software needed to make it work. CMOS chips will be cut out by plasma etching too, so that when bump-bonded to sensor chips, they will not project beyond the sensor edges. During the second stage of the project we will fabricate a 4x4 array of modules, packaged so they overlap in a shingle pattern, with the sensors shielding the readout electronics behind. We will test chip-mounting schemes that use a cylindrical mounting surface, but tilt each sensor to point to the crystal, minimizing parallax. During this second stage of the project, we will package this modular array and demonstrate the feasibility of this kind of modular design. Each silicon sensor will cover 0.96 cm2. When fully scaled up, our process can make sensor/readout modules for less than 100/cm2 dollars in the front end and less than 200/cm2 dollars overall, so this technology is inherently less expensive than CCDs. The detector will discriminate individual X rays yet respond linearly to radiation levels approaching 400,000/s/pixel. Its point- response will remain within a single pixel. Sensitivity will be highly uniform, no distortion will occur within each sensor, and the detector will not exhibit any dark signal or dark noise.